Home · Business · Admin & Office · Procurement

PARTLY

As of 13 August 2026, AI can only partly reduce your business procurement costs.

Most people should hand this to a purpose-built tool.

Can you do it?

15 minutesto a draft.

1 hourto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ita colleague

What the alternative costsNo priced human procurement alternative is provided in the supplied tools data.

If this goes wrong: you pursue an apparent saving that reduces quality, breaches a contract or disrupts an important supply.

What to actually do

  1. Use a tool built for this

    The route this page recommends

  2. Do it yourself

    Second choice

    A chat interface, chat-fluent skill, and roughly 1 hour until you can act on the result.

    How to actually do it

    1. Open your accounting, purchasing or banking system and export the relevant purchase lines with supplier, item or category, quantity, date, price and cost centre.
    2. Gather current supplier contracts, renewal terms, price lists, minimum order rules, delivery terms and any internal quality, security or continuity requirements.
    3. Remove unnecessary personal data, then paste the anonymised exports and relevant document extracts into a new chat with the prompt above.
    4. Add a short description of your business, the goods or services that cannot be disrupted, and any known purchasing constraints in the business context slot.
    5. Ask the model to separate verified findings from hypotheses, then compare every reported total and supplier term with the original export or document before accepting an action.
    6. Give the prioritised actions to the budget holder or procurement lead, obtain approval, and use the supplier questions to request written quotes or revised terms without committing to a change.
    7. After receiving supplier responses, paste the comparable quotes and requirements into the chat and ask for a like-for-like comparison that preserves delivery, quality, security and continuity requirements.

    Prompt

    Act as a procurement analyst for a UK business. Analyse the spend, invoice, contract and supplier information I provide below to find realistic ways to reduce procurement costs. Use only the supplied information and clearly label anything that is missing, uncertain or based on an assumption. Do not invent prices, savings, supplier terms, contract dates, stock levels or service requirements.
    
    Produce:
    1. A table of spend by supplier, category and period, showing the source for each figure.
    2. Duplicate, fragmented or unusually expensive purchasing patterns that merit investigation.
    3. Potential actions, such as consolidating orders, changing specifications, renegotiating terms, removing unused services or seeking competing quotes.
    4. For each action, list the evidence, the information still needed, the operational risks, any contract or supplier dependency, and the person who should approve it.
    5. A prioritised action plan with effort, likely business impact described without invented figures, and a safe next step.
    6. Questions to ask each relevant supplier before changing anything.
    
    Separate verified findings from hypotheses. Do not recommend changing a supplier solely because its price appears lower. Preserve quality, delivery, continuity, security, environmental and compliance requirements stated in the material. Flag any recommendation that needs procurement, legal, finance or operational review.
    
    Business context:
    [describe the business, sites, requirements and constraints]
    
    Data and documents:
    [paste anonymised spend exports, invoice lines, supplier quotes, contract extracts and purchasing policies]

    Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

  3. Hand it to a person

    The distant third

    A person who owns the outcome does this end to end, worth it when the failure is dear.

What it gets wrong

  • AI cannot see purchases, supplier portals, unpublished discounts or contract terms that you do not provide.
  • It cannot tell from a low unit price whether quality, delivery reliability, switching effort or continuity make the change worthwhile.
  • It cannot negotiate with suppliers, obtain binding quotes or accept contractual responsibility for a purchasing decision.
  • It treats incomplete or inconsistent spend data as an analysis problem, not as proof that a saving exists.
  • It cannot approve a change for your business or know which operational risk your organisation is willing to carry.

What caps this at PARTLY: private data access, judgement under ambiguity and stakes of error.

How we scored this

Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.

AxisScore (0–2)
Output2
Inputs1
Verification1
Liability1
Effort delta2
Total7 / 10

FAQ

Can AI really reduce procurement costs?
Partly. It can organise spend data, reveal duplicate purchasing and prepare supplier comparison or negotiation material, but your team must confirm the data and decide whether a proposed saving is operationally safe.
Can ChatGPT analyse my business spending?
Yes, if you provide usable exports, invoices and relevant contract terms. It can find patterns and draft actions, but it cannot access missing supplier information and you should not treat an apparent saving as verified until it matches the source documents.
Can AI negotiate with my suppliers?
It can draft negotiation emails, prepare questions and compare written offers. It cannot build the supplier relationship, judge how far you can press a negotiation or accept terms on your business's behalf.
Is it safe to upload invoices and contracts to AI?
Only after checking your organisation's data policy and the tool's handling of uploaded material. Remove personal and unnecessary confidential information, and do not upload documents if your business has not approved that tool for the data.

Nearby answers

Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.

The newsletter

AI news, new answers and product picks, straight to your inbox.